Universidad Pontificia Comillas. Madrid (Spain)
March 31st, 2025
Summary:
Large cities face a set of challenges resulting from accelerated urbanization and the centralization of populations in urban environments. Therefore, issues such as traffic congestion, increased pollution levels, and overburdened transportation systems have intensified. These problems, combined with the need to align with the Sustainable Development Goals (SDGs) in Europe, the rise of the sharing economy, and the concept of Mobility as a Service (MaaS), have driven the emergence of new forms of urban transportation.
In this context, Bike-Sharing Systems (BSS) have emerged as an innovative, popular, and successful solution, becoming a key component in promoting sustainable urban transportation.
The growth of BSS has not only motivated their implementation in an increasing number of cities but also attracted the attention of the scientific community starting in 2010. This initiated a period of contributions and research across various fields of knowledge. This diversity, however, led to terminological fragmentation, complicating the review of BSS-related articles.
To address the issue of terminological fragmentation, this thesis developed a thesaurus to standardize concepts related to BSS. Additionally, a systematic methodologywas designed to classify BSS-related articles based on their themes and timelines. This classification allowed for the identification of the rebalancing problem.as the primary challenge in this field. Furthermore, it highlighted the need to analyze the historical evolution of these systems, including a detailed review of the BiciMad implementation case. The first part of this thesis, which presents the results of the first article, focuses on these fundamental aspects and lays the groundwork for the rest of the research.
The rebalancing problem consists of two interconnected issues: the first focuses on predicting service demand, and the second on optimizing the process of distributing bicycles across stations to meet this demand.
Addressing both issues requires advanced models to predict and manage the demand for these services.
In the second part of this thesis, presented in the second article, a probabilistic machine learning framework is introduced to model and simulate demand in BSS, with a focus on the BiciMad implementation in Madrid.
The proposed approach uses empirical trip data from 2018 and 2019 to fit and validate theoretical probability distributions, including Gamma distributions for travel distances and Negative Binomial distributions for the number of trips. Unlike traditional regression-based methods, the model described in this thesis incorporates stochasticity and uncertainty, enabling the simulation of alternative scenarios and counterfactual analyses (e.g., “What if...?”). The modeling framework also integrates external variables, such asweather conditions and day types (workdays, weekends, and holidays), to improve accuracy.
Key applications of the model include validating empirical datasets, simulating counterfactual scenarios, and identifying unmet demand due to system limitations, such as the lack of bicycles or docking spaces.
Keywords: Sustainable Mobility; Mobility as a Service (MaaS); Bike-Sharing Systems (BSS); Rebalancing problema; Probabilistic machine learning; Counterfactual analyses;
Citation:
C.M. Vallez Fernández (2025), Modelado Probabilístico y Análisis de la Demanda en Sistemas de Bicicletas Compartidas: Caso de Estudio BiciMad. Madrid (Spain).